Home Theater Geeks 85: IPTV Update

netcasts you love

from people you trust

bandwidth for home theater geeks is

provided by cashfly at cachefly.com

this is home theater geeks with scott

wilkinson recorded october 31st 2011

episode 85 iptv update

this episode of home theater geeks is

brought to you by ford featuring voice

activated sync app link

now you can control select smartphone

apps with your voice so you keep your

hands on the wheel and your eyes on the

road check it out in the 2012 ford

fiesta and at ford.com

technology

and by netflix watch thousands of tv

episodes and movies on your pc mac ipad

iphone or tv instantly all stream

directly to you saving you time money

and hassle for your free 30-day trial go

to netflix.com

twit

hey there scott wilkinson here online

editor of hometheater.com

this week's guest geek is drew major the

cto of ecostar advanced technologies

which is currently working on

iptv a hot topic these days so i'm

really happy to have drew on the show

hey drew welcome

welcome thank you

those who are tuned into the live video

stream at live.twit.tv

or logged into the chat room at

irc.twit.tv can post questions for drew

and i'll pass along as many as i can as

we uh

as we go along here

so uh drew i wanted to start with a

little bit of your history you actually

were one of the founders of novell which

is a well-known networking company and

you developed netware

uh as an application to share fire files

over local area networks so i got that

right

yep it's a long time ago

30 years now

wow wow

and yet uh sharing files over the over

local area networks within a home or a

building or something is now obviously

very common practice

but sharing documents is one thing

streaming media particularly video is

quite another did you did you work on

that in the early days or were you just

mostly concerned with getting a document

from one place to another

well the the networks weren't even

capable of doing that in those days and

so uh

the big deal you got to remember uh we

first shared a 10

megabyte hard drive

i remember my first hard drive was like

a 20 meg and that seemed like a lot at

the time

yeah and so uh you know and that cost

two or three thousand dollars so there'd

be a lot of sense to put a network in

share the files uh

instead of doing using floppy drives

which was the alternative in the early

days otherwise known as sneakernet

yep

but today of course we have we have

enough bandwidth

on local area networks to in fact push

audio and video

around somebody's home

um so the question for me always is

how do you assure

what's called quality of service or qos

when when transporting these these uh

video and audio streams around so they

don't get broken up by somebody sending

a file or checking email someplace else

on the network is that something you

work on

well yeah that's actually the

fundamental challenge facing video

delivery over the internet

if you back up and

or or even in the home under any local

area network if you think about

you know the transition of you know why

networks

work the way they do uh one thing that

was very fundamental early on was

the design of uh

delivery protocols that could deal with

you know everything goes in packets they

could deal with loss packets and so the

invention of ways of dealing with that

the principal way that ended up

happening was a protocol called tcp

which is of course used today

in overall web most most most traffic

and it's designed to deal in a uh

elegant way or at least a pragmatic way

to make sure that the data is fully

delivered

but when they design tcp which by the

way was designed by router vendors to

make sure their routers

didn't melt down originally uh nearby's

uh

what it what it what it does is it makes

sure the data gets delivered but it

hasn't but it imposes no guarantee about

when it does get delivered

and so

the reason which is not that's not a

problem with word processor files but it

is a problem with video files

yeah well it but it also means

something's you know even with a web it

tells you whether i can move my file in

five seconds or two seconds uh depends

on

what's called packet loss and latency

right and uh those two factors happen

and of course you've got to remember the

internet is amazing that it actually

works

disparate set of literally millions of

networks all combined together

and at every point where there's a

router or a switch

even at a modem there's a point where

there can be a congestion point and

the way

they originally designed the routers and

that now is is standard is a there's an

algorithm called random early discard

which basically

throws if if there's a congestion point

if they're trying to put too much data

down a particular pipe

the

router just gives up very quickly and

starts throwing away data

and then what happens is tcp

uh realizes that and backs off and slows

down and of course

because of that the internet works which

is great but the other reason the other

problem the problem with that is that it

doesn't necessarily you can't guarantee

timeliness of delivery and tcp

guarantees delivery but not timeless

delivery and so

even within a house today you know

networks are fast whatever if you're

running wi-fi for example when you set

this thing up to me you told me hey i

can't use wi-fi to connect to skype

well why because wi-fi

typically is susceptible to errors and

stuff and you'll have

a

unacceptable level of packet loss that

would in turn

cause at the videos or our video

conference here wouldn't work

uh well just as a packet loss and so it

you know

even though networks are faster and

better the reality is

in going wireless and in when all your

neighbors are streaming video

uh suddenly

there's a switch point in comcast or

whoever you're using for broadband that

overloads and

data's dropped tcp sessions go slower

video hiccups

and the internet works but

again with video video pays a price in

terms of quality

rich c in the chat room asks how much

buffering goes on to get these delivered

packets

it depends on what you're trying to do

um

let me just describe

what

i invented um

i'll give you maybe just a little more

history this is again this is a geek

show so you can

totally geeky get geeky on me okay so

again i tell ya you know okay so the

internet throws away packets tcp gets

rid of it

uh people realized for many years for 20

years

i got interested in delivering video or

internet in uh mid 90s

or even earlier uh yeah it was mid 90s

because

you know made a lot of sense uh but

the problem was is again how to deal

with this lost packet

thing that there's congestion

some of the packets would be lost

how you dealt with it and of course with

video what they decided to do

is you can decrease the quality of the

video and if there's a if there's

congestion like for example if your pipe

at home your dsl connection is only

one half megabits there's no way you can

put a two megabit

video stream to it you have to figure

out that you have to do less than one

and a half megabit because that's the

size of that

that pipe and so they came up with

algorithms and there was a protocol

defined called rtp rtsp that became a

standard

that basically it was a client server

protocol

where the server would send data

we just start sending packets that have

the video to the client the client on

the other side

would

occasionally reply back saying oh i'm

losing stuff i'm missing stuff and then

the server would look at that and it

would say

i i guess i can't send it at two

megabits i've gotta i better back it off

to one and a half megabits

and so there would be kind of this

interaction where we dynamically

trying to push as much data

through the internet as possible to the

client and then rely on the client to

tell it oh i'm kind of in trouble

i better back off

now that's again was networking

up through really

well today probably half the video

streaming still uses that

they adult primarily most adobe flash if

you go to hulu

it'll be using the rtp rtsp

implementation by adobe

a lot of other streaming still kind of

does that

the problem with that way was though

that it was

that approach was that the server

required intelligence on both sides and

the server was

was kind of blind you know the client

knew what was happening and the server

on the other end had to say well do i

really believe that the client's in

trouble you know

i guess he hasn't squawked enough yet so

i won't

but eventually

you know because because that works

thing can come out of order they can be

delayed you don't know fully what you

can believe because you're kind of at

the other end of a pipe and someone's

shouting at the other end and

it kind of just

it's going through all these sections

and things all get distorted as it goes

through the internet so

you don't fully know it because you

don't want to overreact but you don't

want to under react

so anyway that was the standard way of

doing it um

then the problem was it had two that

that approach had two problems one that

it was based on top of a protocol a

layer actually below tcp called udp

uh that basically

is not a guaranteed delivery thing the

problem with that was that it wouldn't

go through firewalls and people at home

so so in the old days if you did you

like use real networks what would

typically happen there was an early

pioneer of some of this video and audio

streaming in fact for both both of them

they would

they would try to do it with udp

and then if it didn't work they'd back

off and have to use tcp

and

that of course

meant that it

took a while to start

and you know

you push you want to watch video and

it's taking 15 seconds before it figures

out what it can do and what it can't do

because again udp was the firewalls and

stuff over time were that was one way

hackers could you know it's denial

service is like a udp attack and so

there's also protections

and so

basically

that kind of killed or significantly

limited rtp rtsp and so in the end they

put it on top of tcp

and they ran it through

typically the http socket which is the

websocket which was for sure going to be

open for everyone

but in the process i'm sorry i'm sorry

you said the http socket

yeah which is uh which is socket 80

which is your web

your website all your web servers use

that right you know an hdp

uh socket so so you wouldn't have

problems with the firewalls but the

problem at that point was you're on top

of tcp and tcp now has this this other

behavior of

of screwing up the timing of packet

delivery

so anyway that's kind of where the state

of the art

and it wasn't really working that well

uh in in in the past and

what happened was because it was so

complex and stuff some people and i

think most most uh

famously by youtube they decided

well hey we don't need a special

protocol to deliver

uh video let's just make it as a big

file and we'll just use the normal

file down transform that's part of the

web http protocol

we'll just use that file transfer

protocol

and then we'll do a thing called

progressive download which means

we'll open the file start reading the so

the videos on a file we'll open the file

start reading the file we'll see how

fast it's coming and if

if it's coming fast enough we'll

immediately start streaming and hope

that the little progress bar never

catches up to the display bar

or

uh and if that happens then we'll just

pause

okay

right and uh

and we'll just use file transfer

protocol forget these specialized video

protocols they don't really work

by the way file transfer protocols are

easy

and i can just use generic web

infrastructure existing web publishing

infrastructure i don't have to deploy a

special file server or special video

server for the microsoft or the real or

the quicktime or the adobe

it's a lot simpler i'll just use common

web

infrastructure so

that's what youtube did they were very

successful for small videos it was fine

i don't remember in the old days quite

often it would it would hiccup nowadays

it doesn't much because the internet

gets faster and faster over time

so anyway that was sort of the state of

the art and

i had started a company wanting to do

video

distribution i i had to put this on the

record i i had a very interesting now

you know now with steve jobs

i had a very interesting interaction

with steve

back in uh 99 or 2000 oh yeah

i was trying

um i'd met steve because his next

company they wanted to sell it to novell

and

i kept meeting with him every year and i

couldn't figure out how to do it and of

course it made more sense for him to go

back to apple but

he didn't really

know what to do so he kept

meeting with me because he he wanted to

sell next and and again i couldn't

figure out how to make that work and

obviously

it only really worked when he went back

to apple but

right

anyway so i i was in the middle of i was

trying to build content delivery

networks

uh

for video delivery realizing that you

needed big

http web service and all sorts of stuff

and

uh to make that and and the video

delivery would be a you know the next

big thing on the on the internet and so

the problem was at the time the pipes

weren't big enough to do live streaming

so i thought well perhaps

a model could be as people

purchase a thing and then it downloads

it and then maybe

you know an hour two later they can

watch the movie

well that makes some that makes sort of

more sense to me in many respects

because

um you don't know whether it's going to

hiccup in the stream or whatever but on

the other hand it poses the problem of

of copy protection and if you're going

to download something and store it

locally

then you have it locally and how are you

how are the studios going to uh

um address their paranoia

so so anyway you know that they had to

do a lot of you know you have to

obviously do a lot of encryption to to

avoid that problem and a number of

companies have showed up cinema now

others they showed up in you know 2003

2004 et cetera i was i was wanting to

get started a little earlier um and so i

thought well you know who's the guy who

could would get this really quick and i

thought well

steve jobs would and man he owns pixar

and you know he's back at apple and

you know he would he would get it so i

arranged to go meet with them and again

this was

99 or 2000 and

at this meeting and i proposed this idea

to steve and i said you know

steve you know the internet's fast

enough we could start delivering

video movies over the internet what do

you think about that idea

and steve immediately said that's a

lousy idea it's not going to work

and i said because and he said because

because

people want to watch videos immediately

they don't want to wait

and i said well steve you know i knew he

was getting into being a father at the

time i said steve you know when you rent

a video from blockbuster for your kids

do you uh

you go rent it and you can keep it for

five days you rent it a day or two later

you watch it right

and steve came back and he said

nope nope i go reddit and i immediately

watch it

and i

and i

and i was with phil schiller who's i

think his you know his marketing guy he

was the other guy in the meeting and

phil rolled his eyes you know okay this

is steve you know

okay you know steve that's fine you know

whatever you don't and then you know

then we talked about it he had it was

actually a day of earnings call and of

course this is 2000 when apple was just

turning around he was you know he hadn't

even done the ipad pod yet

i'm sure he was working on at the time

so anyway

the discussion ended that steve

kind of didn't want to do a download he

wanted to have

he thought that yate had it immediately

quickly and so

anyway i had that discussion um

literally three years later

which i'll describe here in a minute i

invented a way

to cheaply

do

immediate video streaming

and kind of solve this problem and i

always have the discussion back with

steve you know hey steve you were right

people aren't gonna wait guess what i

invented

the way to to do it over the internet

immediately

and the ironic thing is of course you

know

what does apple do today apple is the

king of downloads

and to wait you know you download patch

you download the thing

and then on the other hand you know i

was promoting that to him he ends up

commercializing it and in the meantime i

invented

what he pushed me for so you know

talk about irony yeah that's really

something we're kind of king of of

both of those camps and of course yeah

anyway let me let me uh

explain now

how i believe video

is going to ultimately be

delivered everywhere and how how

that's going to uh uh

change ultimately the way it you know is

going to eventually get mostly over the

internet and it'll deal with issues in

the house deal with wi-fi issues all

these different issues

which is and this is the invention that

you came up with uh some three years

after that meeting with steve jobs yeah

it was 2003 i kept thinking that people

would fix video streaming and they

weren't and they had this rtp rtsp and

then

then youtube is prior to when they were

acquired to google you know they were

doing

progressive download and it was like

someone's got to figure out how to do it

and there's there's a simple way and

i've been thinking about it on and off

for 10 years but i didn't think i could

was in a position to do it and then one

day i realized that i could do it

or that that

that that it needed to be solved and i

knew how to do it so i invented

it's really an evolution of what's

called uh which was called progressive

download the thing that uh like youtube

is using it

but instead of dealing with video as a

giant file i said well let's

chunk it into different chunks let's

have two second chunks and you gl you

deliver it

together and then

have the client

encoded at different bit rates so

you know you can have a high you know hd

and uh sd and a low bit rate and you

know maybe do you know half a dozen or

even more bit rates

and you mean encode the entire video

file

in multiple bit rates

and have them all available

yeah and then also chunk it into two

second chunks

okay

key with doing the tooth check and

chunks was in that and then access it

via

the web protocols this http protocol

and you just have an intelligent client

that basically says okay give me the

next two seconds it goes and gets it

then it says hey did it take me two

seconds to get this or did it take me

one and a half seconds

well hey if it only took me one and a

half seconds to get this two second

chunk of video

maybe i can get a higher bit rate next

time

so

i go up on the bit rates and i get a

bigger file

and then

things are happening there's errors

wi-fi is having problems or some a bunch

of a bunch of your neighbors are

downloading files and suddenly you can't

get the same bit rate you had before and

suddenly i asked for something it takes

three seconds

then i want to back off and so i better

get a lower bit rate now because i i'm

in trouble but the fact that we chunked

the video

and did it at multiple bit rates and

then had a client that was very

intelligent

they could look and real-time monitor

what was happening and it was all done

in the client

the server was actually just a dumb file

server

dumb you know it didn't have to so it

wasn't a client server architecture

other than the server

just delivers the bits but it had no

intelligence around controlling or

managing

what uh

what was happening and so you put all

intelligence in the client you use the

common

protocols that the protocols of the web

http you do it just for different bit

rates

and

uh

so this is a variation or progression of

progressive low but now i get the best

of both

and so

and of course when you come up with a

good idea

it not only solves the problem you're

solving but you start solving a whole

bunch of other things it's like

this actually worked just as easy just

as good for video on demand as it did

for live streaming

and oh

for live streaming i get time shifting

for free

wow

how do you mean you'd have to store this

that's all you have to do see then

basically in fact maybe even how you're

doing this thing here you you'll do a

live and then you'll encode the vod

thing that's available afterwards

that's right that's exactly right do it

live whatever with our system it's one

and the same

there's no difference

it's really

clean and simple and cheaper

and everything and

so

and it actually works well at higher

qualities so we built that

2003

actually filed for patent in 2005.

we ended up winning in 2005 2006 to do

all the video streaming uh vlog content

for abc for fox for cw for espn we did

the nfl

so all these places are are using this

this technology now right right yeah

that they they are they were using it in

the company that i found it called move

networks and invented

that that i invented this this but it

does require a an intelligent client so

uh

set top box or whatever people have in

their homes

have to have this intelligence built in

right that's right

but the beauty of this approach is

because it's using web and and the thing

and it makes basically it allows enables

the devices streaming to go over an

imperfect network

and deals with it appropriately at the

end point

so you know it's easy to put it on a

phone it's easy to go wireless over you

know 3g it's easy to go over wi-fi

you know anything that has web which is

universal as long as you

install client you can

you can you can stream to it and it's no

longer

uh

hard

uh you know it's the barriers aren't

aren't there in terms of deployability

now

so so so that then let me just finish

the story here

sure

we showed apple disney disney liked us a

lot disney uh they invested in us they

kept telling

apple that they had to support our

protocol so we go up to apple we started

in i think 2007 to go up to apple and

and kept telling them they need to

support this and they'd say they would

then they didn't blah blah blah and then

much to our surprise in 2009

apple proposed a

a standard they called hls http live

streaming that looked really close to

what we'd done and they did it as an itf

standard

was that a patent infringement

oh yeah

we're not sure what we're going to do

about that

okay

copy the same thing so um

that's a separate thing we didn't have

the patent by then though

but anyway um

but the the positive thing that apple

did there was propose it as a standard

and i i view it as kind of a subset of

what we've done uh they uh they

because it was open

we probably should have done it sooner

we didn't

unfortunately we we could have probably

done the same thing anyway so now

there's a standard that's the adaptive

rate standard that's out there with hls

it looks actually pretty good and quite

frankly we're going to start supporting

it because it's kind of a setup we've

done already and i don't really want to

fight that battle in terms of

you know because our negative was we had

to get a client there say

right ended up having about 100 million

with it between abc and fox and stuff we

had about 100 million clients out there

but it was still

hard you know people accept they have

they always give a pause when they have

to install something new so

right and hls doesn't require a

client

uh they do require a client still

but they're being embedded different

places for example you know the only way

to get on an iphone or ipad is to use

hls

so you know you notice you know jobs

block flash there

and certainly they have

soft stuff this so basically because of

that move and again it's a subset we can

still do some of our secret sauce on top

of hls as much as on top of our own uh

writer program so we're transitioning to

hls we publicly announced that

going forward

again what we're going to do with the

patent is a different thing it sounds

like you got to have patents anyway to

be defense

yeah what

it's doing with android

right well listen we've got some

questions in the chat room that i'd like

to get to in just a moment if you hold

on just just a second

i would like to take this opportunity to

thank one of our sponsors for this show

which is uh ford featuring available

voice activated

sync app link

something new here from ford sync app

link enables you to control select apps

from your smartphone

with simple voice commands while you

keep your hands on the wheel and your

eyes on the road very important there

you can do things like voice control

pandora in your car listen to your

tweets with open beak just by asking

with a simple voice command

uh once your smartphone is linked to

pandora for example you can use voice

commands to select your favorite station

make a new one bookmark songs to

purchase give songs a thumbs up or

thumbs down and summon some of the voice

commands include playstation classic

rock bookmark song thumbs up or thumbs

down

and with open beak for example you can

simply say read timelines or read

replies

sync app link is built

on an api platform that allows ford to

continue to work with developers in the

app community to bring apps to life with

voice in their vehicles

voice activated sync app link is

available in the 2012 ford fiesta you

can learn more about it and other

technologies ford is bringing to

vehicles at ford.com

technology

so uh

drew we got some questions in the chat

room or some comments here i just wanted

to uh

bring us up to speed a little bit on

okay

uh which is let me just look for him

here real quick um

does this give you equal quality on both

ends live versus stored media

yes it does yes it does it can be done

well one thing that's nice about

chunking video into two second chunks

and encoding it that way is you can do

a much more uh quality in code on the

two second chunks in live you have he

adds a little bit of latency a little

delay but not not too much you can like

do dual pass

vbr and code on

live which uh we told some people

they said it's impossible you can't do

it it's like you know

yeah we just did

yep

yeah uh uh

dar ck dark maybe uh what software is

this implemented in uh what what'd you

write it in

uh most of it's c plus plus

c plus plus uh again it's a client that

the intelligence is in the client that

installs

um

and it's just a plug-in basically a

browser plug-in

yeah f-loop asks the move player came

and went is this protocol just

implemented directly in flash or

silverlight now you mentioned flash

earlier yeah well both what happened is

microsoft apple and adobe all figured

out that adaptive player the adaptive

bitrate approach using hdp is superior

to what they had done before and so

you know microsoft came up with this

thing called smooth stream hd mic i

can't remember what apple calls apple

calls there's hls

flash calls there's uh uh something else

i don't know the name of it is

but they but basically they're all

evolving towards

this and uh significant amount of

streaming hulu isn't there yet i don't

think

uh netflix dreamer 77 dd asks about

netflix

netflix definitely is adaptive rate they

started we had number of discussions

with them they ultimately decided to do

their own implementation

um i'd argue the one we did is a little

better but flash is good you know

netflix is good enough

right

adaptive right it's definitely adaptive

right

um let's see uh

uh

ans asks couldn't you reverse engineer

it for video upload like skype

i'm not quite sure i understand the

question do you

well

yeah

adaptive freight works best on

on something where it's not not you know

totally real time you know skype here

has to do something different because

you don't want you know you're if you're

talking you don't want more than what a

quarter of a second or whatever

of delay between it and an adaptive rate

isn't suited for that but for video

delivery even live if you can accept 15

20 30 seconds delay off of live

and for scalability you know skype works

great

well i guess you know you guys are you

guys are multicasting too i'm not

exactly sure what what how skype does

that part but but it's definitely

oriented to the live live whereas

adaptive rate does good on live that can

be delayed just a little bit and

definitely for all squad

you mentioned earlier

about

internet speeds in general increasing

and this was one of the questions i

wanted to ask you because

as far as i can tell at least in the

research i've done in the united states

the per capita

internet bandwidth available to uh homes

is not very high compared to some places

uh elsewhere in the world i think it's

well under 10 megabits more like seven

or eight um

so i

i wonder this seems to me obviously to

be uh

one solution to that problem

uh which is getting high quality to the

home

um but with that kind of speed typically

and even less in many cases

the client would have to choose lower

bit rates and you wouldn't get the

quality that you might want

well

thing you have to remember is even if

your

last

mile is is high

these players everywhere even in the

throughout the world they always they

always over subscribe it

sometimes you know 20 50 100 to 1 you

know so so you've got people that think

they all have 10 megabits and there's

100 people with 10 megabits

which you know should be uh

you know 100 000 megabits but the pipe

that actually goes into the internet is

only uh

30 megabits

so you know as everyone and that's that

being the challenge there when you're

starting to talk about more and more

video happening the negative about video

is you watch it a lot with the web you

know you're you know you're you're in

and out you're doing stuff you're you're

just you're bursting and quitting but

with video you know particularly if

you're starting to talk about doing uh

you know a a pay tv thing over the

internet

uh you know you're talking

it's on for five ten hours a day

and so yeah what happens is even if

you're it it's it's the sum of the least

parts you know where they can there's

going to be congestion points up the

path

that are going to really mess you up

even if you've got tons of bandwidth in

your home and tons of bandwidth

uh from comcast or from your fiber to

the home or whatever

certainly on

one of the recent shows that was just on

rerun

here before our show

leo laporte and john dvorak were talking

about

i think leo mentioned something about

how

netflix

accounts for

netflix streaming accounts for 30

of all

internet traffic

and uh when you count everything it's

it's

far more than that so

uh we are certainly moving it seems

towards a model where uh video is being

delivered over the internet and using

quite

a bit of that resource and and then

you've got to recognize that so if

they're taking 30 of the internet and

probably people are watching netflix an

hour to a day

and only maybe 20 percent are netflix

subscribers or 30 percent

then you start seeing that there's going

to be real problems scaling it farther

you know hours a day

and everyone else now

uh again adaptive rate is very suited to

to deal with that uh that

that issue and there's a number of other

techniques that can happen that'll uh

that could solve some of that but

definitely the issue does become some of

it ends up being business model uh

exactly it's not technical it's business

you talk to telcos and they hate netflix

and because netflix

uh does a couple of things they they

they take most of the bandwidth at a

pipe plus netflix even publishes saying

hey this telco is good and this telco's

bad and if you're having problems with

netflix it's not my fault it's you've

got the wrong isp change isps

no wonder they hate it yeah you don't

like that because it's uh it's kind of

uh

a negative thing um

so there's interesting business you know

what incense those people to

you know to uh they have to sometimes

have significant capex to upgrade their

infrastructure

right hundreds of millions of dollars

and netflix

you know takes advantage of that uh in

behalf of the user obviously the user

wants it but

you know the whole the whole

business model and this is where net

neutrality comes in there's just a lot

of

interesting dynamics as to who can

charge what who can force what and who

can

can do what and all that sort of up in

the air i do believe adaptive rates a

better way of choosing it i think that

there's there's there's stuff beyond

adaptive rate that uh that that we're

working on that could

solve this problem somewhat but it

definitely ends up being an economic

problem of how to scale

right and and you are bumping up against

the the obvious limit of you have no

control over

uh the infrastructure the the bottom

line infrastructure of bandwidth into

individual homes that depends on you

know whether they put in copper or fiber

fiber optic or so on uh clearly a

verizon fios has a lot more bandwidth

available to it than most other

systems but still even upstream they

that you know they aggregate it way down

and

the question is and uh you know akamai

uses i'm sorry netflix uses c content

distribution networks like akamai and

they put cat they put proxy caches http

proxy caches that end up being

replicators that fan out the content so

they get one copy into the server send

it out to 100 people you know so it's a

fan out point

well one thing that needs to happen is

the telcos and the isps need to put

those caches further into their network

that would solve some of the problem but

again what's the business model that

that motivates them to do that they're

they're thinking

about that and i'm thinking a lot about

it because it's again that's kind of the

scalability issue

still we're facing in terms of being

able to video

i wonder if the scalability issue might

also apply to the data caps that most

isps are now putting on their their

users

exactly the same thing and how to get

around that and yeah it's all kind of

mixed up together the reason they're

doing data caps is because

you know if they don't do it

their business model doesn't work

they can't charge enough to the customer

to pay for the infrastructure yeah

yeah

has to work it has to work the problem

that i'm faced with always i get back to

is

i'm primarily concerned about quality

and so i wonder if iptv

can ever hope to compete quality wise

with

broadcast cable satellite

which

some would argue isn't necessarily that

good but probably generally speaking

it's better than currently delivered

iptv

not to mention blu-ray

yeah well

i happen to believe that we will get

there uh let me tell you let me tell you

why and how it's

high level um

let me let me first of all explain one

issue uh with with adaptive rate

delivery um

the challenge

that uh adaptive rate has is it's

delivered uh one way of describing is

unicast it's a it's a single stream to a

single user

okay

whereas broadcast is multicast where you

have a single source

whether it's satellite it's just being

get down from one satellite transponder

uh or certainly broadcast it's from an

antenna

uh if it's or it's through the internet

it's still brought it's still broadcast

as a single stream and then

fanned out the problem with broadcast is

again you can't time shift it everyone

has to see everything at the same way

it's a traditional

tv broadcast model and more and more

people want to

do things on their own and they want to

push you do have you do have dvrs of

course now

yeah that's kind of expensive and it's

only it really only works on a set-top

box and now i want to have it on i want

to have the same thing on my iphone and

ipad and

kind of it's kind of complicated so but

the fundamental problem again with with

the ip

uh delivery of adaptive rate is that it

is unicast and so

one solution is to put

these cash replicate proxy caches in the

network close to the user

where again a single stream comes to

that cache it gets stored and then it

gets fanned out to dozens or hundreds of

people

there and so

so in doing that you replicate

somewhat the characteristics of

broadcast in terms of scalability

because that's that really

ends up being the problem if you can

create a business model that incense the

telcos to put caches there

potentially then

you can

caches at the econ at the edge which is

close to the user it's the place where

the video will be originated

if you're in your home you'll talk to

that cash and that cash if the date

isn't there it'll get it from the

internet it'll make one copy but it'll

store it and then if the next second

person asks for it it won't have to go

back to the internet anymore it'll have

it so it'll have it's on its local hard

drive

course that'll require the that'll

require the provider to to install these

caches

which is going to be some measure of

capital expenditure

yeah so you have to give them a business

model to do that and i think we're

figuring out how how to do that but that

in sense will give you the scalability

but it also give you the best of both

worlds in terms of on-demand being able

to

watch everything

and that's kind of quite frankly the

business model with move now we're

pursuing now that we've been acquired by

dish network ecostar

you know you kind of think it's kind of

funny why did uh dish acquire an ip tv

delivery

at company

top realize that the future is going to

be more and more on ip delivery and

instead of

instead of watching that potentially

start cannibalizing the dish delivery

system which i don't think

will take a while to have that happen

instead he wanted to say well let's bet

on the future and so

that's why purchase move

and it's interesting working now for him

because that you know dish obviously has

a lot of clout

and understands the tv delivery and

we're now building a next generation of

that that'll be

ipp based and right now we are selling

it to the telcos

we were just at the telco tv thing

customers uh

but yeah you need a business model that

has sends them to put that

infrastructure in

and i would argue that netflix doesn't

have a as a stronghold business model

there than uh than potentially we would

not that we're gonna compete against

netflix we're

we're gonna we want to empower

telcos to do iptv similar to what

verizon's already done with

technology so uh qwerty in the chat room

asks how far out till we can select only

the channels we want on dishnet

so in other words a la carte um

systems rather than you know

packages

i would love to have that happen i'll

tell you a couple things that hold that

back the first thing is

when you buy a subscription from cable

you know a pay tv subscription from dish

or directv

or comcast or verizon whoever

it costs them somewhere at least seven

or eight hundred dollars

to to to fully provision that client you

know in the case of dish they have to go

and put a satellite dish they have to

give you a receiver

all and they have to do a truck roll

all that costs money

so if you if you look at it and the

average subscription is for 24 months

they you know they try to lock in at

least for 24 months well why do they do

that because

it cost 700. and if you divide 700 by 24

months

it's like 12 to 15 dollars a month just

to pay back the capital expenditure they

had to make for that equipment

oh they have they have this huge burden

they have to pay to to to to even start

breaking even because they they invested

a lot to give you that dish and then

that receiver and so

that tends to make it so they want to

buy sell you bigger packages because

they have to recoup all this capital

expenditure

okay

yeah it could change that a little bit

hopefully it would

the challenge there is the content guys

also like to sell all their packages and

and

they've all figured out that the best

way is to have big packages that are

expensive than everyone makes money

and if they break it up perhaps they'll

make less money

never a good thing got it you know who

who wants to go do the deal if you're at

disney or fox or turner do the deal that

two years ago two two years later made

it so that you were making uh

80 of what you used to make you know

right right that's a deal now the

challenge i think they're all working

worrying about is cord cutting

and they're worrying about uh you know

netflix you know man instead of paying

70 bucks a month for a cable i can maybe

uh maybe netflix is good enough for 10

bucks a month so

it's a very interesting time with

business models and there's and

disruptive things are happening and i

think adaptation and what we're doing

here

potentially can accelerate those things

certainly what netflix is doing uh even

with their recent challenges you know

their value is pretty high

but

we're in the middle of transition

i would love to have the alucard thing

as i'm pushing for that i think with

over the top the probability of that

happening

is going to be high but it's it's

definitely business model issues

yeah well certainly one of the um

one of the companies in the middle of

this transition

is another of our sponsors for the show

that i'd like very much to thank for

their sponsorship and that of course as

you mentioned earlier is netflix which

streams thousands of tv episodes and

movies directly to your

pc or your tv

your mac even the ipad or iphone

uh you can uh and some android phones

for that matter if you have a gaming

console you can stream from the wii or

the ps3

uh

watching netflix directly on your tv if

you have a roku box same thing you can

get netflix just about everywhere

and from just about any device

and you can watch these tv shows and

movies instantly

without any hassle or returning discs or

anything like that

you can even begin watching on one

device and

finish watching on a different device

whichever way you choose to net to

access netflix you can watch as many

movies and tv shows as you want and you

can cancel any time

for twit listeners there is a free

30-day trial that you can get by going

to netflix.com

twit

and be sure to use this url when you

sign up for that free trial

netflix.com slash twit and we thank

netflix for their support of the twit

network and we hope you enjoy

watching instantly with netflix

um

so drew one of the uh one of the things

that

i i recently i've been thinking a lot

about is the coming well first of all

let me let me just say 3d

has been a pretty big thing lately and i

was wondering if your adaptive uh bit

rate system has uh any capability to do

uh 3d

uh streaming

yeah

all 3d is is just you're sending more

data the two different sides of the

thing and again our adaptive

system being so dynamic can actually

reliably deliver 3d

uh better than others you know it's much

more flexible it's just a small it's a

software thing it's not this bigger

upgrade and so yeah i would argue i

would think that that eventually

everything goes adapter rate and

certainly 3d will be one thing that will

go on

what about 4k that was the other my

other question i've recently seen some

demos of 4k that look awfully good and

and to these people who all say oh

physical media is dead you know

everything's going to be streamed

uh i say well that may be but are we

going to really be able to get 4k

streamed again it's the size of the bits

and the pipes you know if it you know

again it ends up being business model uh

in my opinion yeah yeah

well if if the

your isp provider is given the right set

of incentives to upgrade and to pay for

all the capital expenditures he's got to

do to to to make the pipes bigger

for you uh then uh all that will happen

certainly verizon right now with fiber

to the home is very capable of doing

that

and then

what att has done and really dsl can

can move there

and and quite frankly the cable guys can

as well it's just a matter of the

business model

everyone has a big enough pipe

at the last mile the real question is is

is is the

stuff

packed back of that and all the costs

associated with that and what's what

motivates them to

you know and again our strategy in

selling our product now through the

telcos is to give them that revenue

stream to motivate them to upgrade their

broadband pipes

i assume that you can you can

accommodate any uh codec

yeah that that we want to use right have

you heard of this new high efficiency

video codec hevc

i've i've talked to people who uh

are building encoders who are

envisioning that they can do better jobs

right now we're you know h.264

right all the progress all that's good

you know that's kind of why i say

ultimately the quality is going to reach

the quality of broadcast is because

every barrier and i've i think i've

described a number of them here

based all those barriers are going to go

away it's going to compress a little bit

better the pipe's going to be

bigger to your home and again adaptive

rate gives you that flexibility

and

i i think the other thing that that i

that i would maybe want to finish with

is just this thought that the thing that

the beauty of adaptive rate ultimately

is that it can tolerate errors in the

delivery path

if you think about how video how

traditional broadcast cable satellite

broadcast video has been delivered it is

it assumes that the entire delivery path

has to be

perfect

you can't lose anything

uh

and so that unfortunately

adds a significant expense

to put that high high degree of

reliability in the entire delivery path

whereas adaptive rate because

the the the reliability is imposed at

the tcp and at the higher level

everything below it

can be significantly simpler

and cheaper

and so ultimately it's like for example

you know even within uh within broadcast

systems today they all use ip in their

internal networks they used to have all

sorts of specialized networks but the ip

ended up being cheaper what they have to

do is is over super over provision it so

that that it never loses any data

but what's happening is the open the the

the internet the ip delivery paths

because

it can be simpler and because adaptive

rate allows those paths allows the

internet to be a little flaky allows the

routers to throw away packets allows

wi-fi to attempt to temporarily lose

some data when someone turns on the

coaster in the middle of uh you know

whatever spike interferes with the wi-fi

temporarily because at a high level

the resiliency is at the high level the

rest of the thing can be all the all the

the delivery path can can have problems

it can be simpler and it doesn't matter

see and so ultimately

then simplicity and price

wins that's why i would predict long

term

that he'll all switch away from uh these

more the old generation having to be

perfect delivery paths

to uh an ip based which doesn't have to

be perfect

well this answers webby's question in

the chat room is streaming ever going to

be going to supplant over the air for

real-time tv

um and i think your answer is yes

eventually and ultimately it will

ultimately and again it's a cost

it's a cost basis thing that's that's

the beauty of the ip the internet

networking

versus the older systems and

and again you're just starting to see

that adaptive break takes advantage of

that

this is why i believe

ultimately most video delivery is going

to be adapted break certainly netflix

believes in it today

microsoft don't be

apple

yeah yeah

we know that's the future it's just a

matter of time and it's a matter of

business model and we'll

and that's gonna happen i think i think

the change on video delivery in the next

uh five years is gonna be

gonna be tremendous we're just getting a

taste of it with netflix and with

streaming to our our different devices

ultimately

all the data that's you know all the

stuff that's on satellite and broadcast

will be available it's just incredible

i've i've have been a skeptic of that of

that uh proposal in the past but you are

starting to turn me into a believer

in the pudding i think that within the

next year you'll see some very

interesting things

well i'm certainly looking forward to it

and uh maybe in the in a year or so

you'll come back and share with us what

you have done between now and then

enjoy too that would be fun

great thank you so much drew major the

cto of echostar advanced technologies a

fascinating conversation and i do hope

you'll come back and see us again thanks

so much

thank you

okay um oh one more question drew sorry

you don't have any website or or place

where people can get more information is

that correct

not really not really

it's too advanced

okay so sorry about that folks

yeah there's a move network stuff that's

kind of old but that will be required by

echo star

so right we don't have a new option

when when you do have more information

online please let me know so that i can

pass it along to my audience

uh

and

whatever i learn you'll certainly read

about at my online home which is

hometheater.com

you can also email me at scott twit dot

tv

and you can follow me on twitter at ht

geek scott

next week my guest geek is scheduled to

be

me

i decided to do

a chat room uh chat so i will be simply

here uh answering questions from the

chat room

and um

something that i like to do once in a

while and uh seems to get a good

response from everybody out there so get

your questions ready

uh anything home theater related is fair

game and uh just uh be ready to be in

the chat room and

post them and i will be looking at as

many and answering and talking about as

many as i can

two weeks from now my guest is going to

be

ace k tsuyuzaki the cto of panasonic

north america

big time guy

he is he's a very big wig in the

consumer electronics industry

and i'm really looking forward to a

conversation with him in two weeks so be

sure to join me for that as well as for

the uh

chatting with the chat room episode next

week

until then

geek out

you